Vet Pathol 29256-260 (1992)
Canine Pericardial Mesothelioma
S. P. MCDONOUGH, AND A. H. TOBIAS
N. J. MACLACHLAN,
Key words: Dogs; mesothelioma; pericardium.
Mesothelioma of the canine pericardium has been de-
scribed infrequently, usually as one manifestation of wide-
spread dissemination of the n e o p l a ~ m This
. ~ ~ ~report
~ ~ de-
scribes pericardial lesions in three dogs that were presented
to the Veterinary Medical Teaching Hospital (University of
California, Davis, CA) because of recurrent pericardial ef-
fusion. All three dogs had been treated with repeated peri-
cardiocentesis and eventually by subtotal pericardiectomy.
Histologic examination of the resected pencardial sacs from
all three dogs revealed nests of mesothelial cells. All three
dogs died within 1 month of surgery, and necropsy of one
dog confirmed fulminant metastatic mesothelioma subse-
quent to pericardiectomy. The other two dogs were unavail-
able for postmortem examination.
The three dogs included an 11-year-old male Airedale Ter-
rier (case No. l), a 9-year-old castrated male Golden Re-
triever (caseNo. 2), and an 11-year-old Airedale Terrier cross
(caseNo. 3). All three dogs had signs ofright-sided congestive
heart failure characterized by ascites and distended jugular
veins. Radiographic examination revealed globular cardiac
silhouettes; examination by electrocardiographyshowed small
complexes and electrical alternans, and echocardiographic
examination confirmed pericardial effusion. No masses as-
sociated with the right atrium, aorta, or other structures with-
in the pericardial space were visualized during the ultrasound
examination. The Airedale Terrier (case No. 1) was treated
with repeated pericardiocentesis over a 4-month period, and
the pericardial effision of the Golden Retriever (case No. 2)
was drained twice over the course of 1 year. The Airedale
Terrier cross (case No. 3) had a single bout of pericardial
effusion in 1985 that recurred 5 years later and was then
drained twice. All these dogs subsequently underwent medial
sternotomy for exploration of the thoracic cavity and subtotal
pericardiectomy. Neoplasms were not detected in any of the
dogs during exploratory thoracotomy. Several weeks after
surgery, all three dogs developed marked pleural effusion.
The owners of the Airedale Terrier (case No. 1) and Airedale
Terrier cross (case No. 3) declined further treatment, and
subsequently both dogs died but were unavailable for nec-
ropsy. The Golden Retriever (case No. 2) was treated with
repeated thoracocentesis and two doses of intracavitary Cis-
platinum@for suspected pleural mesothelioma. The dog re-
sponded poorly to therapy and was euthanatized. At nec-
ropsy, this dog had a moderate serosanguinouspleural effision
Fig. 1. Pericardium; dog No. 3. Nests of apparently neoplastic cells are surrounded by fibrovascular stroma. HE. Bar
= 50 pm.
Fig. 2. Pericardium; dog No. 3. Pancytokeratin (AEUAE3) staining is intense and distributed diffusely throughout the
cytoplasm of apparently neoplastic cells in Fig. 1. Avidin biotin peroxidase complex method, Mayers hematoxylin coun-
terstain. Bar = 100 pm. Inset: Same section stained for vimentin. Note positively stained cells at the arrow. Bar = 10 pm.
256
and a large irregular multilobular mass that was firmly ad-
herent to the diaphragm, epicardium, and the right pleural
surface. Neoplastic nodules were also present in the lung and
tracheobronchial lymph nodes.
A11 tissues collected after pericardiectomy or necropsy were
immersed in 10% neutral buffered formalin and embedded
in paraffin. Sections were cut and stained with hematoxylin
and eosin. The subserosal connective tissue layer of all peri-
cardial sacs contained multiple nodules of large cells arranged
in solid nests or irregular cords and acini that were bounded
by a scant to moderate amount of fibrovascular stroma (Fig.
1). The pericardial sac from the Airedale Terrier (case No.
1) also had multiple nodules of glandular elements that pro-
jected in small dome-like masses above the surface of the
serosal layer. Cells were cuboidal to polygonal with a large
oval nucleus that contained coarsely stippled to vesicular
chromatin and one to two large prominent nucleoli. The cells
had abundant amounts of granular eosinophilic cytoplasm
and distinct cell borders, and one to three mitotic figures
were present per high-power field (500 pm in diameter). Dif-
ferential diagnoses included metastatic squamous cell car-
cinoma, adenocarcinoma, and mesothelioma. The histologic
appearance of cells within neoplastic nodules collected at
necropsy of the Golden Retriever (case No. 2) was identical
to that of cells within the pericardium, but rafts of cells were
frequently present within endothelium-lined vessels.
Ultrastructural analysis and immunohistochemical stain-
ing of the tumor cells were undertaken to confirm their his-
togenic origin. The presence of both cytokeratins and vi-
mentin in tumor cells within the pericardial sacs of affected
dogs was established by avidin biotin peroxidase complex
method staining using commercially available monoclonal
antibodies that were specific for high- and low-molecular-
weight cytokeratins (AE 1/AE3, Boehringer Mannheim, In-
dianapolis, IN), primarily low-molecular-weight cytokera-
tins (PKK1, Labsystems, Raleigh, NC and CAM5.2, Becton
Dickinson, Mountain View, CA), high-molecular-weight cy-
tokeratins (A575, DAKO Corp., Carpinteria, CA), and vi-
mentin (ICN Biochem, Lisle, IL). Factor VIII-related antigen
staining (Calbiochem Corp., San Diego, CA) was done in an
attempt to demonstrate vascular invasion by neoplastic cells.
All staining was done using the standard avidin biotin per-
oxidase complex method of Hsu et al. and the Vectastain
kit (Vector Laboratories, Burlingame, CA). All tests were
Vet Pathol 29:3, 1992 Brief Communications and Case Reports 251

Fig. 3. Pleural mass; dog No. 2. Pancytokeratin (AEUAE3) staining is intense in neoplastic cells. Avidin biotin peroxidase
complex method, Mayers hematoxylin counterstain. Bar = 50 pm.
Fig. 4. Pleural mass; dog No. 2. Vimentin staining of the same tissue as in Fig. 3. Note clusters of intensely staining
cells at the arrows. Avidin biotin peroxidase complex method, Mayers hematoxylin counterstain. Bar = 50 pm.
Fig. 5. Electron micrograph. Pericardium; dog No. I. Tumor cell in subserosal nest. Note microvilli around the entire
circumference of the cell, numerous well-developed desmosomes (arrow), and prominent intercellular spaces. Bar = 2 pm.
Fig. 6. Electron micrograph. Pleural mass; dog No. 2. Note well-developed microvilli with primary and secondary
branching. Bar = 1 pm.
258 Brief Communications and Case Reports Vet Pathol293, 1992

Table 1. Summary of immunohistochemical staining of neoplasms.

Cytokeratins
Tumor Type Case No. Breed Sex* Age AE 1/AE3
(years)
Intensityt Pattern$
Pericardial 1 Airedale Terrier M 11 +++ Peripheral
mesotheliomas 2 Golden Retriever M/c 9
biopsy material +++ Diffuse
necropsy material ++ Diffise
3 Airedale Terrier cross M 11 +++ Diffuse
Pulmonary 4 Cockapoo M/C 16 ++ Peripheral
adenocarcinomas 5 Doberman Pinscher F/s > 16 +++ Peripheral
6 Shetland Sheepdog M 12 +++ Peripheral
Epithelial type 7 Yorkshire Terrier F 13 +++ Diffuse
mesotheliomas 8 German Shepherd Dog M/C 14 +++ Diffuse
9 German Shepherd Dog M/c 15 ++ Diffise
10 Labrador Retriever M 7 +++ Diffuse
11 Terrier cross F/s 10 + Diffuse
Chemodectomas 11 12 Labrador Retriever F/s > 16 0
13 Australian Shepherd Dog F/s 11 0
14 Golden Retriever F/s 14 0
* M = intact male; M/c = castrated male; F/s= spayed female.
t ++ + = most cells stained with many staining intensely; + + = many cells stained moderately and some stained intensely; + = many
*
cells stained weakly and some stained moderately or intensely; 0 = no cells stained.
Diffuse = the reaction product was evenly distributed throughout the cytoplasm; peripheral = the reaction product was confined to the
periphery of the cytoplasm.
9 ND = not done.
11 Chemodectomas were weakly positive for Chromogranin A but negative for neuron-specific enolase.

performed on formalin-fixed paraffin-embedded tissue with
3,3-diaminobenzidine (Sigma, St. Louis, MO) as the chro-
mogen. All cytokeratin-stained slides were predigested with
0.1% trypsin (Sigma, St. Louis, MO) for 20 minutes before
the primary antibody was applied. The staining pattern of
the cells within the affected pericardial sacs was compared
with histologically confirmed cases of canine pulmonary ad-
enocarcinoma (n = 3), disseminated epithelial-type meso-
thelioma (n = 5), and chemodectoma (n = 3), as summarized
in Table 1. The pericardial sac tumors and the chemodec-
tomas were further evaluated with Chromogranin A (Incstar,
Stillwater, MN) and neuron-specific enolase (DAKO Corp.,
Santa Barbara, CA). Negative technique controls were per-
formed by substituting the primary antibody with phosphate-
buffered saline and also by using normal rabbit serum. Pos-
itive and negative tissue controls consisted of a block of 2 1
normal canine tissues.
Tumor cells within the affected pericardial sacs coex-
pressed vimentin as well as both high- and low-molecular-
weight cytokeratins that typically were distributed diffusely
throughout the cytoplasm (Fig. 2). Neoplastic cells in tissues
taken at necropsy from the Golden Retriever (case No. 2)
had the same staining reactivity (Figs. 3, 4). The epithelial-
type mesotheliomas usually had a similar staining pattern,
except vimentin staining was restricted to scattered cells.
Whereas the epithelial-type mesotheliomas inconsistently re-
acted to CAM5.2, both pericardial sac tumors on which this
staining was done reacted moderately to intensely. None of
the pulmonary adenocarcinomas expressed vimentin but did
react moderately to intensely with both high- and low-mo-
lecular-weight cytokeratins. The chemodectomas were neg-
ative for both cytokeratins and vimentin, weakly positive for
Chromogranin A, and negative for neuron-specific enolase
(data not shown). The staining pattern of the adenocarci-
nomas was distinctive, with a markedly enhanced peripheral
(membrane) region imparting a ring-like appearance to the
cells. This staining pattern was not seen to an equivalent
extent in any of the other types of neoplasms.
Selected thin sections of affected pericardial sacs and met-
astatic tumors collected at necropsy from each case were
stained with 2%uranyl acetate in 95% alcohol for 16 minutes
and in Reynolds lead citrate for 9 minutes and examined
with a Zeiss transmission electron microscope. Ultrastruc-
tural details of the neoplastic cells were distinct (Figs. 5, 6).
Numerous long microvilli with primary and secondary
branching surrounded the entire circumference of the cells.
Adjacent neoplastic cells were joined by frequent well-de-
veloped desmosomes, and intercellular spaces were promi-
nent. Cells had abundant cytoplasm that contained numerous
bundles of tonofilaments arranged circumferentially around
the nucleus, rough endoplasmic reticulum, scattered glycogen
granules, and moderate numbers of mitochondria. Sections
removed at surgery from the pericardium of the Golden Re-
triever (case No. 2) also contained cells with numerous mi-
crovilli, but autolysis prevented their definitive character-
ization. The ultrastructural characteristics of neoplastic cells
collected from metastatic tumors at necropsy from this dog
were compatible with those of mesothelioma.
Vet Pathol 29:3, 1992 Brief Communications and Case Reports 259

Table 1. Extended.

Cytokeratins
Vimentin
PKKl CAM5.2 HMW Intensity
Intensity Pattern Intensity Pattern Intensity Pattern
+++ Peripheral ++ Peripheral +++ Diffuse +
++ Peripheral NDO +++ Diffise +
+++ Diffise ND +++ Diffuse ++
+++ Diffuse ++ Diffuse +++ Diffuse +
++ Peripheral 0 ND 0
+++ Peripheral ++ Peripheral ND 0
+++ Peripheral ++ Peripheral ND 0
+++ Diffuse + Diffuse ND 0
+++ Diffuse 0 ND +
+++ Diffuse 0 ND +
+++ Diffuse + Diffuse ND +
+++ Diffuse 0 ND +
0 ND ND 0
0 ND ND 0
0 ND ND 0

Pericardial effusion is an infrequent cause of cardiovas-
cular disease in the dog.' Reported etiologies include neo-
plasia, benign idiopathic pericardial effusion, primary car-
diac disease, trauma, infection, and uremia. The three dogs
described in this report had a distinctive syndrome of peri-
cardial effusion associated with mesothelial cell proliferation,
which in one case progressed to disseminated mesothelioma.
Because of similarities in the postsurgical outcomes, we be-
lieve the mesothelial proliferations in the two remaining dogs
also indicated mesothelioma, although their malignant po-
tential could not be confirmed because they were not nec-
ropsied. No direct evidence of lymphatic invasion was found
by examination with light microscopy or electron microscopy
or by the use of Factor VIII-related antigen staining to iden-
tify lymphatic vessels. The pericardial effision was most like-
ly secondary to increased hydrostatic pressure from lym-
phatic obstruction. The rapid postsurgical deterioration in
all three dogs likely resulted fiom transthoracic spread of
neoplastic mesothelial cells at the time of subtotal pericar-
diectomy.
Immunohistochemical staining clearly identified these tu-
mors as mesotheliomas and not carcinomas. Normal serosal
tissues consist of a surface mesothelium derived from mul-
tipotential subserosal cells in the underlying layer of vascu-
larized connective Resting multipotential subserosal
cells express only vimentin but coexpress vimentin and low-
molecular-weight cytokeratins as they proliferate. As these
cells differentiateinto surface mesothelium, they may express
high-molecular-weight cytokeratins but cease to express vi-
m e n t h 6 Although expression of cytokeratins and vimentin
by mesotheliomas is variable, often reflecting the histologic
type of tumor, immunohistochemical examination is increas-
ingly being used to diagnose mesotheliomas.24JoThe peri-
cardial mesotheliomas in this study consistently expressed
vimentin and both low- and high-molecular-weight cytoker-
atins. These cells may have been derived from multipotential
subserosal cells, but the possibility of derivation from the
surface mesothelium with dedifferentiation cannot be ex-
cluded. The ultrastructural features of the neoplastic cells
also were characteristic of mesothelial rigi in.^.^.^,^.'^
Acknowledgements
We thank D. Naydan for her enthusiastic assistance with
the immunohistochemistry and Drs. C. McCullough, D.
Danilenko, and B. Rideout for their original descriptions of
the biopsy material. We also thank Dr. R. J. Higgins for his
helpful discussions.
References
1 Berg RJ, Wingfield W Pericardial effision in the dog: a
review of 42 cases. J Am Hosp Assoc 20:721-730, 1984
2 Bolen JW, Hammar SP, McNutt MA: Reactive and neo-
plastic serosal tissue; a light-microscopic,ultrastructural,
and immunocytochemical study. Am J Surg Pathol 10:
34-47, 1986
3 Cibas ES, Corson JM, Pinkus GS: The distinction of
adenocarcinoma from malignant mesothelioma in cell
blocks of effusions: the role of mucin histochemistry and
immunohistochemical assessment of carcinoembryonic
antigen, keratin proteins, epithelial membrane antigen,
and milk fat globule-derived antigen. Hum Pathol 18:
67-74, 1987
4 Corson JM: Pathology of malignant mesothelioma. In:
Asbestos-related Malignancy, ed. Antman and Aisner,
pp. 179-199. Grune and Stratton, Inc., Orlando, FL,
1986
5 Harbison ML, Godleski JJ: Malignant mesothelioma in
urban dogs. Vet Pathol 20531-540, 1983
6 Holt JP: The normal pericardium. Am J Cardiol 26:
455-465, 1970
260 Brief Communication.s and Case Reports
7 Hsu SM, Raine L, Fanger H: The use of antiavidin an-
tibody and avidin-biotin-peroxidase complex in im-
munoperoxidase technics. Am J Clin Path01758 16-82 1,
1981
8 Ikede BO, Zubaidy A, Gill C W Pericardial mesotheli-
oma with cardiac tamponade in a dog. Vet Pathol 17:
496-499, 1980
9 Trigo FJ, Morrison WB, Breeze R G An ultrastructural
study of canine mesothelioma. J Comp Pathol 91:53 1-
537, 1981
Vet Pathol29260-262 (1992)
Desmin and Vimentin Immunocharacterization of Feline Muscle Tumors
J. MART~N J. H. Vos, AND F. N. VANMIL
DE LAS MULAS,
Key words: Cats; desmin; feline tumors; immunohistochemistry; intermediate filaments.
The immunohistochemical identification of different in-
termediate filaments is a useful aid in the histogenetic clas-
sification of undifferentiated tumors because the intermedi-
ate filament profile of mammalian cells is rather tissue
specificloand is generally preserved during neoplastic trans-
f ~ r m a t i o n Some
.~ of the commercially available antisera to
intermediate filament proteins can be used on formaldehyde-
fixed, paraffin-embedded material,I3 allowing retrospective
studies to be performed on routinely processed tissue spec-
imens. Desmin is an intermediate filament protein present
in cardiac and skeletal muscle and also in smooth muscle of
viscera and v e s ~ e l s . ~The
J ~ immunohistochemical demon-
stration of desmin in tumor cells is a reliable method to
elucidate the myogenic character of soft tissue tumors in
human beings.12 Rhabdomyosarcomas have histologically
variable histomorphologic appearances8and consequently can
be difficult to diagnose, particularly in the absence of distinct
histomorphologic features indicating rhabdomyoblastic dif-
ferentiation. In human beings, desmin approaches absolute
sensitivity and specificity as a marker for muscle cell differ-
entiation in rhabdomyosarcomas of the embryonal and al-
veolar types and is also a sensitive marker for pleomorphic
rhabdomyosarcoma.l* Desmin labeling of rhabdomyosar-
comas in several other species has been reported previously:
in dogs,1.6rats,ll and mice.5 In cats, the myogenic character
of a presumed rhabdomyosarcoma could not be demonstrat-
ed immunohistochemically by labeling for desmin.
Seven feline soft tissue tumors histologically diagnosed as
rhabdomyosarcomas were typed immunohistochemically for
desmin, vimentin, glial fibrillary acidic protein, and neuro-
filament proteins. All cats were European short hairs. The
tumors were at different locations, i.e., at the larynx, mam-
mary glands M5 (inguinalkM4 (lower abdominal), the ear
region, and the thoracioabdominal wall and diaphragm, with
metastases to the lung and kidney. The same intermediate
filament profile was also analyzed in histologically normal
260
Vet Pathol 293, 1992
10 Wick MR, Loy M, Mills SE, Legier JF, Manivel JC:
Malignant epithelioid mesothelioma versus peripheral
pulmonary adenocarcinoma: a histochemical, ultrastruc-
tural, and immunohistologic study of 103 cases. Hum
Pathol 21:759-766, 1990
Request reprints from Dr. S. P. McDonough, Department of
Veterinary Pathology, School of Veterinary Medicine, Uni-
versity of California, Davis, CA 95616 (USA).
feline tissues (skeletal muscle, intestinal smooth muscle, ten-
don, and spinal cord) and in 24 histologically nonmuscle
feline soft tissue tumors: neurofibrosarcomas, fibrosarcomas,
lymphosarcomas, and undifferentiated sarcomas (Table 1).
Tissue samples had been fixed in 10% neutral buffered for-
malin for 24 to 36 hours and embedded in paraffin. Com-
mercially available polyclonal antisera directed against
chicken gizzard desmin, calf lens vimentin, human spinal
cord glial fibrillary acidic protein, and a monoclonal anti-
serum raised against human brain neurofilament proteins
were obtained from Euro-Diagnostics B.V. (Apeldoorn, The
Netherlands). The peroxidase anti-peroxidase and indirect
immunoperoxidase techniques were applied to polyclonal
and monoclonal antisera, respectively, as described else-
where.13 All primary antibodies were incubated overnight at
4 C and diluted 1 :80 (polyclonals) and 1 : 10 (monoclonal)
in phosphate-buffered saline, pH 7.6.13 Normal skeletal and
smooth muscle tissue sections were used as positive controls.
Negative controls included the substitution of the specific
primary antibodies by both phosphate-buffered saline and
normal rabbit or mouse sera diluted 1 :80 or 1 : 10, respec-
tively. Counterstaining of immunoperoxidase-treated sec-
tions was performed with Hams hematoxylin; tissue sec-
tions were incubated for 1 minute.
Histologically, all seven rhabdomyosarcomas showed in-
filtrative growth, although two appeared to be partially en-
capsulated. All tumors contained sheets or nests of various
sizes of highly pleomorphic round polyhedral to spindle-
shaped cells, with eosinophilic cytoplasm and a broad range
in the nucleus: cytoplasm ratio (Fig. 1). Densely packed in-
terlacing bundles of fusiform cells with eosinophilic cyto-
plasm and blunt-ended uniform nuclei sometimes predom-
inated. A variable number of round to angular pleomorphic
cells with either one or two bizarre nuclei or several smaller
ones was present in all cases. Striations were not seen after
phosphotungstic hematoxylin staining, mitoses were not par-